THE EXTERNAL STRUCTURE OF THE LEPIDOPTEROUS LARVA. 53 



moving, therefore, it is helpless, and must travel from place to place 

 with as much speed as may be possible. Every observer knows that 

 the tree-feeding larvae of the other groups mentioned above are 

 extremely slow in their movements. It is essential, above all things, 

 that a tree-feeding larva should hold very firmly, and this it is enabled 

 to do by spinning silken threads and ladders, and by the possession of 

 remarkably strong and well-developed prolegs. The large Saturniids, 

 arboreal Sphingicls, Lasiocampids, etc., cling with amazing tenacity, 

 but, at the same time, they walk with extreme slowness. With them, the 

 opening and closing of their prolegs is a remarkably complex operation, 

 in which a whole army of muscles is brought into play. The 

 Geometrid larva has to cling as tightly as these. At the same time it 

 has to move more rapidly, hence it has reduced its prolegs to the 

 smallest possible effective number, and has, especially, anal ones of the 

 very best kind. Thus it is able to obtain a long stretch for each step, 

 and is able to progress with comparative speed. The young Noctuid 

 larva, too, has often a considerable amount of travelling to do in 

 search of food (eggs being often laid away from the food-plant, etc., 

 ante, p. 13), and a certain amount of looping increases its activity by 

 lengthening the step ; and this is, perhaps, much more important in 

 the young state when the larvae have an arboreal habit. It may be, 

 therefore, that rapidity and facility of progression is a great part of the 

 object in view. An Arctiid larva, when travelling rapidly, hardly uses 

 the prolegs at all, but progresses by a rapid looping movement, the 

 ordinary progression, segment by segment, being altogether too slow 

 for its needs. 



Every field naturalist has observed how a Geometrid larva will 

 maintain its hold upon a twig and eat a leaf, and, for this, reach is 

 also required. The difference between the way in which a tree- 

 feeding Geometrid larva and a Sphingid larva will attack a leaf is 

 remarkable. The Geometrid stretches itself out to its full length, and 

 eats as much as it can reach without moving, often beginning near 

 the tip and devouring the whole leaf. The powerful Sphingid larva 

 pulls the leaf towards itself, and thus does by greater strength what 

 the Geometrid larva does by greater reach. 



The Geometrid form, therefore, appears to be correlated with habits 

 of (1) greater reaching or stretching power, (2) greater speed. It 

 does not seem to have any important phylogenetic significance. 



In a previous part of this chapter, we have referred to the fact 

 that lepidopterous larvae have a certain number of ocelli on each cheek. 

 Landois considers that these do not essentially differ from compound 

 eyes, and states that if many of them were grouped together they 

 could hardly be distinguished from compound eyes. In each ocellus, 

 he says, the cornea is divided into three lenses, each corresponding to 

 three nerves, each with a separate terminal enlargement, forming the 

 so-called crystalline bodies. Each ocellus, therefore, might be re- 

 garded as being, in reality, composed of three. On the other hand, 

 the three arches of the cornea are so closely connected together, that 

 they give the impression of forming a simple cornea. The three lenses 

 are also very closely pressed, and the three nerves unite into one. 

 Under these circumstances, Landois regards the ocelli of caterpillars 

 as a connecting link between simple and compound eyes, and proposes 

 for them the name of " ocelli compositi." Chapman says : That 



